Recent technological advances in high-throughput experimental analysis enable modern biologists to[unreadable] collect data at the genome scale, and use them to decipher biological principles and uncover the elements[unreadable] of complex behaviors in living organisms. The advances and changes in the research paradigm require[unreadable] development of a set of tools that biologists need for scientific discovery. Among these, computational and[unreadable] data analysis tools are essential, and are largely provided by the fields of data mining, bioinformatics and[unreadable] statistics. We propose to introduce a new approach to functional genomics studies, and hypothesize that[unreadable] the global expression profile of any organism could provide a universal phenotype for direct prediction of[unreadable] biological function. We will develop a set of computational tools to treat such phenotypes, perform the[unreadable] corresponding data analyses and infer predictive functional models from experimental data. Our efforts will[unreadable] be based on an arsenal of state-of-the-art data mining approaches. We will adapt existing tools and[unreadable] develop new ones to help us infer reliable predictions, to find what biological changes took place following[unreadable] environmental or genetic change, and to explain the relevant biological background. Our methods will infer[unreadable] interactions between global expression profiles, mutant fitness, gene function and annotation, and classical[unreadable] biological phenotypes, such as chemotaxis, morphogenesis and development. Using correlation studies,[unreadable] the methods will decompose expression profiles to biologically meaningful components that will enable us[unreadable] to reason on the functional changes and their relations at the genome scale. Most importantly, we will test[unreadable] and adjust these tools in collaboration with the biological projects, ensuring their practical utility. We will[unreadable] package our methods into open-source toolboxes, using component-based design and a visual[unreadable] programming paradigm to make the tools accessible to users that are not programmers or computer[unreadable] experts. We will make that package freely available to the research community. This project will also define[unreadable] and maintain the information infrastructure for the entire program, and will provide databases that will store[unreadable] experimental information and related data on hundreds of mutants. Finally, we will develop server-based[unreadable] software to provide public access to the vast amounts of data produced by this program and to selected[unreadable] data analysis tools through the world wide web.